What Is a Layer 2 Rollup?

If you have ever paid a high fee to make a simple transfer during a busy period, you have felt the core limitation of a Layer 1 blockchain: block space is finite. A Layer 2 rollup is the most widely adopted answer to that problem. It is a separate network that does the heavy work of processing transactions off the main chain, then posts a compressed record back to Layer 1 so the base chain can keep it honest. The result is much lower fees without giving up the security that makes the base chain worth using.

This explainer covers what a rollup is, how the two major designs—optimistic and zk—differ, and where the real risks live. For the underlying mechanics of how transactions settle, see how a transaction gets confirmed.

Why Layer 1 needs help

A Layer 1 (L1) such as Ethereum runs every transaction on every full node and stores the results forever. That redundancy is exactly what makes it trustworthy—and exactly what limits its throughput. When demand exceeds the available block space, users bid up fees to get included. You cannot simply raise the limit without asking more of every node, which would push out smaller operators and reduce decentralization. So the scaling strategy became: keep L1 lean and secure, and do the bulk of execution on a second layer.

What a rollup actually does

A rollup executes transactions off-chain, bundles hundreds or thousands of them together, compresses them, and posts the resulting data to L1. The name comes from this bundling—it “rolls up” many transactions into one L1 submission. Crucially, the transaction data is published to the base layer. That is what separates a genuine rollup from a sidechain: because the data is on L1, anyone can independently reconstruct the rollup’s state and verify that its operators followed the rules. The rollup borrows the base chain’s security instead of inventing its own.

Ethereum.org frames the fee impact directly, contrasting an average transaction on Ethereum with a far cheaper average on Ethereum-backed networks. The savings come from many users sharing the cost of a single L1 settlement, plus data-compression techniques and, more recently, a dedicated cheap data format called blobs.

The role of the sequencer

Most rollups today use a sequencer: an operator that receives users’ transactions, orders them, executes them off-chain, and posts batches to L1. Sequencers give rollups their fast, cheap user experience, but early rollups run a single sequencer, which is a centralization point. A sequencer can, in the short term, reorder or delay transactions. It cannot, however, forge invalid state that the base layer would accept—because the correctness of the batch is enforced by one of two proof systems.

Optimistic rollups: trust, then verify

An optimistic rollup is optimistic in a literal sense: it assumes every batch posted to L1 is valid and does not publish a proof of validity up front. Instead, it opens a challenge period. During that window, any watcher who spots an invalid state transition can submit a fraud proof. Ethereum’s optimistic rollups use interactive, multi-round proving that narrows a dispute down to a single computational step, which an L1 contract can then evaluate cheaply to decide who is right.

The trade-off is time. Because a fraud proof can arrive late, withdrawals back to Ethereum must wait for the challenge window to close—roughly seven days. Everyday transactions inside the rollup feel instant; it is only the final exit to L1 that carries the delay. Third parties sometimes offer faster “exits” by fronting liquidity, but that is an added service with its own trust assumptions.

Zk rollups: prove first, then settle

A zk rollup (zero-knowledge, also called a validity rollup) takes the opposite stance. For each batch it generates a cryptographic validity proof—a compact piece of math attesting that the new state was computed correctly from the old one. Ethereum verifies that proof on-chain. If the proof does not check out, the batch is rejected. There is nothing to challenge after the fact, so there is no multi-day fraud window; withdrawals can finalize as soon as the proof is verified.

The cost of that guarantee is engineering complexity. Generating validity proofs is computationally intensive, and building a proving system that fully matches the base chain’s programming environment has historically been harder than building an optimistic rollup. That is one reason optimistic rollups saw earlier, broader adoption—though zk technology has matured quickly.

Optimistic vs zk at a glance

Why fees drop so much

It is worth being concrete about where the savings come from, because “cheaper” can sound like marketing. Three effects stack. First, amortization: hundreds or thousands of users share the single L1 cost of settling a batch, so each user pays a fraction. Second, compression: rollups strip and pack transaction data far more efficiently than raw L1 transactions, shrinking what must be posted. Third, cheap data lanes: the introduction of blobs gave rollups a dedicated, lower-cost place to publish their data instead of competing for the same expensive space as ordinary transactions. Together these turn what might be a dollar-scale fee on the base layer into a fraction of a cent on a busy rollup. The trade-off is that you are now trusting the rollup’s additional machinery, which is exactly why the risk checks below matter.

Data availability: the quiet foundation

A rollup’s security rests on one requirement: the data needed to reconstruct its state must be available to everyone. If an operator could hide transaction data, no one could build a fraud proof or independently verify a validity proof’s inputs. This is why rollups post data to L1. The introduction of blobs—a cheaper, temporary data lane—cut costs sharply; blob data can be pruned from clients after roughly 18 days, which is long enough for anyone to verify or challenge while keeping long-term storage cheap.

Rollups vs sidechains vs validiums

“Layer 2” gets used loosely, so it helps to draw lines. A genuine rollup posts its transaction data to Ethereum, so its security derives from the base layer—anyone can reconstruct and check the state. A sidechain is a separate blockchain with its own consensus and its own validators; it may connect to Ethereum via a bridge, but it does not inherit Ethereum’s security, and if its own validator set fails, Ethereum offers no fallback. A validium resembles a zk rollup—it uses validity proofs—but keeps transaction data off-chain, trading some security and data-availability guarantees for even lower cost. The practical takeaway: the more a system relies on Ethereum for data and settlement, the more of Ethereum’s security it borrows. Marketing labels do not always match this reality, so it is worth checking how a given network actually publishes its data.

What to check before trusting a rollup

Because rollups vary widely in maturity, a few questions separate a robust one from a risky one. Are fraud or validity proofs actually live? Some early “rollups” launched before their proof systems were fully enforced, meaning users temporarily trusted the operator. Who controls the upgrade keys? If a single party or small multisig can instantly change the rules, that is a centralization and security risk regardless of the proof system. How decentralized is the sequencer? A single sequencer can censor or reorder, and if it goes offline, users need a way to force transactions or withdraw directly through L1. Independent trackers publish “stages” of rollup decentralization precisely so users can compare these properties rather than take a project’s word for it.

Where the real risks are

Rollups reduce fees, but they add moving parts you should understand. Sequencer centralization can allow short-term censorship or reordering. Upgrade keys—often a multisig held by the rollup team—can, on less mature rollups, change the system’s rules. And the biggest historical danger is the bridge you use to move funds between layers. Bridges hold large pools of assets and have been among the most exploited components in the entire ecosystem. Prefer a rollup’s official native bridge, verify the URL, and move a small test amount first. Our deep dive on bridge risk explains why cross-chain movement is the weak point.

Rollups are a genuine advance: they let ordinary users transact affordably while still anchoring to a secure base layer. But “backed by Ethereum” is a spectrum, not a switch. The safest approach is to learn how a specific rollup handles its sequencer, its proofs, and its bridge before trusting it with meaningful funds. For the wider context, the how it works hub and our note on blockchain finality tie these ideas together.

Sources

• Layer 2 — ethereum.org
• Optimistic rollups — ethereum.org
• Zero-knowledge rollups — ethereum.org
• Danksharding & data availability — ethereum.org